中文摘要：

在电能质量检测中,去噪和保留突变点信息是两个十分重要的问题。为此提出了一种新的基于梯度倒数加权平均算法的自适应滤波方法。该方法首先对电能质量采样信号点建立五个模板,依据检测点与五个模板均值的梯度变化规律,进行突变点判决。在传统梯度法的基础上,选择最佳模板,并用该模板均值与模板内各点幅值的差值代替传统梯度算法中的梯度值,然后对不同类型的点采用不同的算法去噪。实验结果表明,与传统的梯度倒数加权算法和五点均值滤波法相比,改进的算法能够更好地清除噪声,同时较好地保留突变点信息,有针对性地解决了电能质量检测中的两大重要问题。

英文摘要：

Novel gradient inverse weighting filter was proposed for two crucial issues in power quality detection, which were the de-noising and keeping the break points＇ information of the power quality signal. Firstly, this paper designed five cover templates for each point, and judged the points that were break, according to the variety rule of disparities, which were the denoised point magnitude with the average magnitude of each cover template. Then the de-noised point was processed in one cover template, which average magnitude was most close to the de-noised point magnitude. Calculating the disparity value of the average magnitude with each point in this template, it was efficient to take this disparity value instead of the gradient value in the classical gradient inverse weighting method. However, this novel filter is adaptive to carry on the different de-noising method for the non-break points and the break points. The simulation experiment results indic, ate that, for the effects of de-noising and keeping the brake-points＇ information, this new filter was much better than the classical gradient inverse weighting filter and the average filter, and it effectively resolves the two important problems in power quality detection.